Antisocial behavior is widely recognized as the most powerful predictor of later adjustment problems for males of any childhood behavior. Childhood antisocial behavior predicts a constellation of negative outcomes in adulthood, including criminal behavior, substance abuse, at-risk sexual behavior, relationship failure, low achievement in education, unstable work history, and early fatherhood. The current proposal would extend a longitudinal study which has spanned later childhood and adolescence into the adult years (to ages 26-27 years). The multimethod/multiagent study began with two cohorts of fourth grade boys (total N=206), and now has a sample size of over 200 at ages 20-22 years (a 98% retention rate). The application proposes to examine factors associated with continuity and discontinuity in antisocial behaviors across this developmental span, with particular attention to the social interaction processes involved, as well as contextual factors and the young man's prior behavior. Hypotheses are based on the coercion model of family interaction (Patterson, 1982) the deviancy training model of peer influence (Dishion, Spracklen, Andrews, & Patterson, in press), and the developmental failure model (Capaldi, 1991; 1992;). Hypotheses also incorporate other theoretical approaches to the explanation of stability and change, including Scarr and McCarney's (1983) conceptualization of individual-environmental interaction. The study will involve assessments of adult process predictors and each of the dependent variables listed above for the OYS males. Assessments will include self-report, parental, peer, and co-worker reports, records data (police records, Division of Motor Vehicles records and credit reports), and assessment with a male peer at age 23-24. Funds for assessment of the young man's relationship with an intimate partner are currently being requested in a completing continuation of a separate grant. Longitudinal prediction and development models using late childhood, adolescent, and young adult data will be tested using innovative multivariate techniques. Processes related to behavioral change in young adulthood will be a major focus.